API-Sowter Transformers

[khstudio]

I can't really do a freq plot... don't have the right equipment.

But I do have a B&K scope & a tone generator.
The only problem is my tone Gen is set at 600 ohm's output & can't change it so I'm not sure how accurately it will work when testing 150 ohm inputs???

It's just a cheap hand held model but works pretty well for most jobs & makes a good square wave on the scope.

 

[khstudio]

OK, one more stupid question.

Non of the 6 wires are for the case or core (like the Cinemag has & connects to Ground)
Should the metal case be grounded :?:

If so... I have to rig something up. no big deal.

 

[CJ]

Try it with the can floating and then put an alligator clip on it, with the other end grounded, and see what happens during a listening test with the amp wide open.

600 ohms is fine, because the place we want to measure is way out of the Bass zone, which is where the generators crap out.

Remember, it is a 150 Ohm input at a certain frequency.
1000 Hertz? I forget.


That thing will drive 150 all day at 10 K.

You can even use a regular voltmeter to check for peaks, everything is True RMS nowadays, weather you like it or not, so a hand held could be used for a cheap and dirty test.

 

[khstudio]

Remember, it is a 150 Ohm input at a certain frequency.
1000 Hertz? I forget.

That thing will drive 150 all day at 10 K.

Good Point that I didn't think of :thumb:

So (in general) I can use my Sig Gen to test 1:10 TX's too (secondary loaded with 150k, reflecting 150 to primary) as long as I set the Gen to 10k or above???

Let me know if this is the right way:

Set my Sig Gen to 10k, Square Wave & look for ringing (on the scope)
Then add the ZOBEL & adjust until it looks good?

If Pat reads this he's going to shoot me :razz:
He's the one that told me about the Sowters & said they were his favorites... set up with the Original API 312 Values / Zobel.

I'm just trying to learn & dig a little deeper here. :green:

I emailed Sowter a few times + did a lot of searching & haven't heard or found anything about this, so...

Time for testing & listening.

 

[rodabod]

[quote author="CJ"]You can even use a regular voltmeter to check for peaks, everything is True RMS nowadays, weather you like it or not, so a hand held could be used for a cheap and dirty test.[/quote]

Only thing to watch out here is some moder DMMs have pretty poor frequency response.

[quote author="khstudio"]
So (in general) I can use my Sig Gen to test 1:10 TX's too (secondary loaded with 150k, reflecting 150 to primary) as long as I set the Gen to 10k or above???

Let me know if this is the right way:

Set my Sig Gen to 10k, Square Wave & look for ringing (on the scope)
Then add the ZOBEL & adjust until it looks good?
[/quote]

Pretty much. Get a pen and paper and note what you see. I'd sweep with a sine wave first and accurately measure voltage level from 10K upwards as CJ suggests. Set your volts/division very low so you can gain accuracy and view the tops of the peaks.

Square wave gives you a single tone plus an infinite series of odd harmonics on top which sum to the square wave shape which you see. Look at the edges which is where the HF exists. Big positive kinks indicate too much HF. Ripples indicate HF ringing.

Check it, and if the API zoebel doesn't help, follow the standard Jensen procedure for choosing a network.

 

[CJ]

You can tune the Zobels by ear, also.

Hold R constant, then audition caps.
The hold C constant, and audition resistors.

Then go back and forth til you get the best response with the lowest loss.

The Zobel is not a perfect fix, it sucks a little power from the input signal, so that you lose a little mid range in order to bring the ringing down.

To see this, put an AC current meter on the primary and watch the ma or ua as the Zobel is switched in and out.

When your DMM starts singing, you are probably losing a little signal. :?

 

[khstudio]

Spent some time on this today - testing square waves, sweeps (with them in the pre) & listening, also every test I put the Cinemag 75101 against it to see & here.
Well, If I were to judge everything on just the sweeps alone I'd probably pic the Cinemag but the "Sound" is WAYYYYYYYYY different.
This Sowter just has a nice vibe & seems a lot more "REAL" sounding. While the Cinemag is more "Sterile & Flat" sounding.
At first I thought the Cinemag had more highs to it, then after further listening & going back & forth several more times it started to hit me & become more clear WHAT was going on.
I really like this Sowter... a lot more punch & realizm... I can't stress the Real thing enough.

Now, the Cinemag requires NO Zobel, just a 150k across secondary.
With the Sowter, there was no recommendations (which is why were discussing this).
Raw (No loading or Zobel) the Sowter was "Ringing" with little to no overshoot.
The Cinemag RAW, had NO ringing at all but a decent overshoot... even after adding the recommended 150k it STILL had a small overshoot. :?

Back to the Sowter.
So I tried the API 312 Zobel & it stopped the ringing but seemed to roll too much off... so I called Pat & he said he used a 120pf in place of the 220pf... I tried this & a it eased the high "Bend/roll-off a little & the tiniest amount of ring came back.

I finally said f*#k-it with the testing & put the cards back in my (8 channel) pre... all the other channels use the Cinemag 75101's.

Like I said earlier, this is where the difference became much clearer.
I also just got done sweeping (with my computer - good converters) from 1 Hz - 22 kHz & BOTH "Looked great.

So, the sweep told me they are balanced throughout the usable audio spectrum (of human hearing, digital audio & as much as my Computer app can accurately do) neither with any significant high-end roll-off.

So it all comes down to the "Sound".

The ZOBEL I ended up with after trying several different combination's was 120pf & 5.1k... it seemed to be the best trade off so far.

This may change after I do more listening or get a response from Brian Sowter.

 

[maxwall]

How do you arrive at your rc base values for the zobel to begin with ?
there are infinite combinations, yes.

Could a trimpot be used to simply the resistor side ? Then drop in different caps ....

 

[CJ]

What is the DCR on the Cinemag Sec vs the Sowter Sec.?
Thanks!

Also, you can compare Leakage Inductance to see why one has more of a peak at 40 K.
Just short the Sec leads and measure the Pri. Inductance from both Primaries.

I bet they are different between the Sowter and Cimemag.

Sorry if you already bolted it up, don't bother.

I will see if I have the leakage specs for the orig. API input.

 

[khstudio]

CJ,
The Cinemag:
Primary pin 1 & 3 = 3.19H & 21.5 ohm
Primary pin 2 & 4 = 3.20H & 33.4 ohm

Secondary pin 5 & 8 = OVER 200H & 2.21k
(My Meter doesn't go over 200 Henries!!!)

As far as the Sowter goes, the DCR (from memory) were close but just under from the 2622 spec sheet you posted.

But I think the Primary inductance was higher... like around 8 Henries.
Same thing goes for the output Henries... over 200. as far as my meter goes.

 

[khstudio]

My EMAIL WITH BRIAN SOWTER!!!

Hello,
> I just received a pair of your API 2622 (9820f) input TX's.
>
> I'm hooking them up and need to know about the ZOBEL:
> Should I just use the Original Zobel network from the API 312 mic pre or
> do you recommend other tunning or loading?
>
> Also, a lot of current/modern input transformers (like the 75101 from
> Cinemag) recommend a 150k across the secondary but NO Zobel...
> What do you recommend?
>
> I also have a switch on my pre-amps to have the option to choose between
> the 150 or 600 ohm tap (1:4 or 1:8 approx)
> Just thought I'd mention this in case it affects your recommendations...
> the 1:87 is the most important to me.
>
> Thank you... I'm really looking forward to hearing these.
>
> Kevin

Brian Sowter wrote:
> > Hello Kevin

This is a difficult one. In general our transformers do not need a Zobel
network as you can see from the frequency response curves we publish in the spec sheets where we have them.

Obviously the precise response depends on the source and load impedances, particularly the load capacitance so it is hard to be precise anyway.

For this particular transformer I recommend using just a 100K resistor.
If you are able to measure the overshoot on a square wave or a rising
response at high frequencies it is best to lower the secondary load resistor till it cleans up. The only disadvantage of doing this is a loss of signal as the resistor is lowered. The ideal response is a 6 dB per octave roll of with no overshoot. It is most unlikely that this transformer would need a Zobel (RC) network.

> > Sorry not to be more specific.
> > Best regards

> > Brian Sowter

Thanks Brian,
We've been discussing this on the LAB - Prodogy Forums:
http://www.groupdiy.com/index.php?topic=362783

I've looked everywhere on the Sowter web site & could not find any
frequency response curves or
spec sheets with them???

Also... are you sure 100k? This seems a little low.
The reflected impedance would be pretty low from it to... between 1000 &
1200.

I have spoken with several of my friends on & off the forum that have
set these TX's up so many different ways.
One reason is that there is no definitive answer as to weather or not
these are "Drop in" replacements or
how they should be set up properly listed on your site.

If you don't mind, I'd like to post your responses on the forum to help
others... or you can yourself if you want.

Thanks man... these things do sound GREAT!

Kevin

Hello Kevin

Thanks for this. The reflected input impedance is obviously important.
100K was only a gut feel. We don't have specs for all our transformers on
the website (yet). This is only because of the workload. As there is
obviously a lot of interest in this I will measure one as soon as possible
and let you know the result.


Best regards

Brian Sowter
SOWTER AUDIO TRANSFORMERS

 

[khstudio]

Now I'm really confused... especially after what I measured, then the email from Brian :sad:

I also talked to Pat again last night & he has SEVERAL API pre's ALL set up a little different & the more he thought about it he wasn't sure how he set up the Sowters.

Right at the end of our conversation I got the 1st email from Brian.

I'll tell you... there was definitely "Ringing" going on & it NEEDs something :!:
Brian mentions not needing a Zobel with these & initially recommended a 100k only - but I tried a 150k & it did nothing about the "Ringing" but the ZOBEL did!!!

... me confused. :?

I'm really surprised that this wasn't addressed (from the start) by Sowter... being this is advertised as an API 2622 Replacement.

Hopefully Brian can do the "proper" measurements & tests to set the record straight. I have a feeling some sort of Zobel is needed from what I saw.

 

[rodabod]

Kevin, I have to agree with Brian that this is a complex subject since it does depend partly on the conditions in which you set for the transformer.

For a start, the transformer loading is not just the Zoebel network, but also the load of your line input (this could be say, 10K to 100K) which sits in parallel to your network.

In addition to this, there is a question of taste. Sometimes a little resonance towards the top-end can be nice. Sometimes not.

I'll tell you... there was definitely "Ringing" going on & it NEEDs something

I'm not sure how much ringing we are talking about here, and I'm not convinced that it definitely needs something. There are compromises which occur when you adjust a Zoebel network and it's a balancing act like many aspects of finding an optimum in electronics.

For my last network setup for some Beyers, I had a compromise in allowing a small amount of ringing in order to extend response beyond 20KHz, and also had to allow around a 1dB rise around 16KHz in the process.

I think you need to try a couple of preamps together with different setups (eg. no Zoebel and Zoebel set for minimum ring) and have a listen. A mic splitter would be a fair test source here in my opinion assuming your input trafos and opamps are the same.

Hope this helps.

Roddy

 

[okgb]

I have used the Lundahl 1538 xl , which are said to not need
a zobel as well , not testing but with no percievced problem
In fact when used in my G.R. MP-2 i took out the zobel that was in
for the cinemag and it seems to be a tiny bit more " airy "

Sorry if this is off topic from the a.p.i.'s but another example of xfmr
that doesn't need zobel

 

[maxwall]

Just curious, What kind of audio artifacts are evident with a little bit of ringing and when building a proper functioning zobel network or just a resistor by itself to null out some of the ringing , is it normal to loose db's in certain desirable frequencies while gaining in others that are unwanted. What is the ultimate goal ? is it a compromise between frequency response and nulling out ringing and distortion. Is it trying to keep the frequency response as flat as possible over a broad frequency spectrum while minimizing unwanted ringing and distortion ( unwanted waveform behavior )

can some one explain this from the macro level so newbies can understand at the micro level.

 

[khstudio]

Maxwell,
I'm not ignoring you man... I'm just a little caught up on the Sowter for the API right now.

Is it trying to keep the frequency response as flat as possible over a broad frequency spectrum while minimizing unwanted ringing

I'd have to say yes to this one.

 

[khstudio]

[quote author="rodabod"]Kevin, I have to agree with Brian that this is a complex subject since it does depend partly on the conditions in which you set for the transformer.[/quote]
Understood... but the conditions "should" be known because it's supposed to be a 2622 replacement in the API circuit.

For a start, the transformer loading is not just the Zoebel network, but also the load of your line input (this could be say, 10K to 100K) which sits in parallel to your network.

When you say "Line input" what are you referring to: the "Loading resistor" or the amp itself?
If your talking about the Loading resistor... then I really get confused on this one. A lot of newer designs like the Cinemag, call for one... like 150k for the 1:10 ratio. Reflected impedance (to the mic) would be 1500.
Putting a "10k - 100k" parallel to the Zobel seems a bit low, no???

Also, the 2622 in the API pre ONLY had a ZOBEL... No Loading resistor... which is confusing.
I HAVE listened several times to these trannys (cinemag & Sowter) with & without the load resistor & there's a definite change to the "FEEL" of the pre. With the load, the "sound" is much "tighter" & without = it seems to breath more... seems more natural, open & resonant.

I guess this is were "Personal Taste" comes in.

In addition to this, there is a question of taste. Sometimes a little resonance towards the top-end can be nice. Sometimes not.

I'll tell you... there was definitely "Ringing" going on & it NEEDs something.
Kevin

I'm not sure how much ringing we are talking about here, and I'm not convinced that it definitely needs something. There are compromises which occur when you adjust a Zoebel network and it's a balancing act like many aspects of finding an optimum in electronics.

This is why I asked Brian... I'm not sure how much is too much but compared to other TX's I've testing, there was a LOT!!!

Like I said earlier - I tried a load resistor & it did NOTHING for the ring but the original API 312 Zobel DID = (5.1k & 220pf). But it seemed to roll off a little too much, so I changed the 220pf to 120pf & it seemed to be the best compromise.

For my last network setup for some Beyers, I had a compromise in allowing a small amount of ringing in order to extend response beyond 20KHz, and also had to allow around a 1dB rise around 16KHz in the process.

I did sweep the pre (with the Zobel of 120pf & 5.1k - NO LOAD Res) & it was dead flat from 20Hz - 22K. So I MAY be on the right track & it does sound very good... I've noticed (with ALL input TX's) that when you install a "Loading Resistor" the sound, especially the lower end, "Tightens up"
There may be a little too much resonance, the way I have them set up now, by not having SOME sort of Load Resistor added too... I'm giong to try this now!

[quote author="rodabod"]
I think you need to try a couple of preamps together with different setups (eg. no Zoebel and Zoebel set for minimum ring) and have a listen. A mic splitter would be a fair test source here in my opinion assuming your input trafos and opamps are the same.

Hope this helps.

Roddy[/quote]

Thank You for your help :thumb:

I'm going to do what you just said... listen. :green:

 

[rodabod]

[quote author="khstudio"]
When you say "Line input" what are you referring to: the "Loading resistor" or the amp itself?
If your talking about the Loading resistor... then I really get confused on this one. A lot of newer designs like the Cinemag, call for one... like 150k for the 1:10 ratio. Reflected impedance (to the mic) would be 1500.
Putting a "10k - 100k" parallel to the Zobel seems a bit low, no???[/quote]

Apologies - I thought you were also testing the output transformers as well which would obviously see the line input which follows them (eg. on your sound card).

Also, the 2622 in the API pre ONLY had a ZOBEL... No Loading resistor... which is confusing.
I HAVE listened several times to these trannys (cinemag & Sowter) with & without the load resistor & there's a definite change to the "FEEL" of the pre.

I see what you are saying now regarding there being no regular load resistor. I think at this point though, you are the engineer, and you must decide what is best for you. What might API have done if they chose to use the Sowters?...

 

[CJ]

Kevin, thanks for the inductances.

Try doing it over again, only short the sec leads together.

Also, do not agonize over the input termination too much.
There are a lot of chassis out there that ring like heck and sound good.

A transformer that does not ring at all might actually sound flat compared to say, an ol UTC ring-o-matic A-10.

 

[khstudio]

Hello Kevin

We measured a 9820 frequency response and found a small resonance peak of
0.4 dB at 35 kHz. The roll off after this is quite close to the ideal 6 dB
per octave. The resonance peak is not as good as I would have hoped for but
I would still regard it as a small effect. Even so I would expect a little
overshoot to be measurable with a fast square wave.

It did not make any real difference with a 150 k Ohm load or a 250 k Ohm
load.

As a result of this I would consider a Zobel network and the original 5.1K
and 220 pF in series would be a good start.

By the way I designed this transformer form good measurements on the
original so I would expect our transformer is very close to the original. I
would be interested to know if you can hear or measure any difference with
and without the Zobel.

Best regards

Brian Sowter
SOWTER AUDIO TRANSFORMERS